As previously pointed out, this is a big question. I suggest that you instead focus on getting part of the question answered, so divide your development up onto small stages that address different parts and look to solve the small parts, one at a time.
The one to start on, because I think you can't actually do this on device, is the voice recognition. I am not aware of any way to do voice recognition in 3rd party apps on device. I am not aware of any APIs to do so either. So I suggest you focus on this, see if you can figure out how to do that, and then if you solve that move on.
If you can't figure out to do it, you should search here to and elsewhere to see if someone else has, or there have been similar questions asked. If you still get no where, then can ask a new question on here. I suggest you mark this question as solved and don't extend it because its subject is not specific to your voice recognition question.
